Skip to main content

Execute a skill

POST 
/api/v2/agent/:agentCode/skill/:skillCode/execute

Runs one skill in isolation: no planning, prompt, agent model or conversational pipeline is executed. The skill is validated against the skills declared in the resolved agent version. The request body is optional and must use the same JSON shape the LLM builds for the skill's tool call, which you can discover with the skill schema endpoint.

Request​

Path Parameters

    agentCode stringrequired

    Code of the agent that declares the skill.

    skillCode stringrequired

    Code of the skill to execute.

Query Parameters

    connectionId uuid

    Optional personal connection to authenticate the skill with. Only valid for skills configured with a personal connection; when omitted the caller's default connection for the connector is used.

    culture string

    Use this param to override the culture of the response. Options: - en (default) - es

Body

Optional tool-call arguments as a JSON object.

    any

Responses​

Returns the skill execution result, or the pending actions the caller must complete before it can run.

Schema

    content stringnullable
    jsonContent nullable

    pendingActions

    object[]

    nullable

  • Array [

    • oneOf

    type stringnullable
    url stringnullable
    connectorId uuid
    connectorName stringnullable
    connectorImgUrl stringnullable
    type stringnullable

  • ]

Loading...